Skip to main content

Designing Best-Fit Classes with the Class Placement Engine

(This post is by Ben Hacking, Deputy Principal at the Vienna International School)

Established in 1978, the Vienna International School (VIS) is a CIS accredited, IB World School situated in Vienna, Austria.  The School serves students from kindergarten to grade 12 who come mostly from the diplomatic corps and international businesses based in Vienna.  Diversity is one of VIS’s greatest strengths and the School is proud that of the 1368 students attending, approximately 112 nationalities are represented around 85 native languages are spoken altogether.  This diversity is also represented amongst our approximately 270 faculty members, leadership teams and administrative staff. 

VIS joined the Learning Analytics Collaborative (LAC) in 2017-18 after an extensive internal review of, and reflection on, our own data for learning practices.  Over time, we have worked with the LAC to develop a number of engines to support and empower our faculty with the data tools and inquiry processes they need to support student learning.  One such engine was the Class Placement Engine.

THE CHALLENGE
We know that in order for students to learn, they first need to feel safe, supported and a sense of belonging. For students, this feeling starts with the classrooms they’re placed in. At VIS, the goal for the class placement process is that all students are placed in agreed classes that will support their academic and social-emotional well-being, and that have the highest potential to become safe, caring and cohesive learning communities. In doing so, faculty are asked to create balanced (or intentionally imbalanced) classes across a range of criteria such as age, gender, academic ability, behaviour and more. In addition, students should end up with positive social and learning partners while avoiding negative ones. 

For a time, faculty generated paper baseball cards with key demographic, academic and social information on them and teams would then gather to move the cards around on a table in different combinations. While this was helpful, we found that there were simply too many variables to consider in meeting the goal of placements and in the end, there were still many students who were unhappy with their placements and classes that could have been better balanced. Given all the data we were collecting and working with for this process, we approached the LAC to help us develop an engine that give us a starting point for class mixing.

THE SOLUTION
The Class Placement Engine is a set of tools that utilize a range of student profile data to support the process of class placement and the analysis of class/cohort compositions, distributions and social connections. Similar to how a school might use physical “baseball cards”, the Class Placement Engine allows teachers to manually move and group digital versions of such cards into the desired number of class sections while seeing the data points on the composition of class sections change in real time.  

Because managing so many variables was a challenge, an algorithm was designed that gave teachers a starting point for class mixes. Teachers simply select the number of sections to mix classes into, press play, and watch the engine generate class mixes automatically based on criteria we set.  The algorithm is designed to achieve balanced sections with the maximum amount of positive partnerships and the minimum amount of negative partnerships per section. Once complete, teachers can click and drag students across sections to make minor changes based on their own professional judgment which an algorithm can never replace. Teachers can then save that version, generate other versions, and export it into a excel or csv format which is helpful for generating class lists later.

THE OUTCOME 
The development of the engine and underlying processes for class mixing has proven to be a big support to our faculty, particularly in that the algorithm provides a reliable starting point for teachers to build on and that teams can generate multiple versions of classes with ease.  What we’ve also found is that there is value to the engine well beyond the mixing process itself. Because the placement engine reveals particular students associated with a data point (e.g. the number of strong readers in the cohort) teachers are able to get to know their students before they walk in the door. 

Likewise, our counsellors, grade and department leaders as well as senior leadership team are able to start gleaning information on the cohort to start planning the curriculum as well as academic, social-emotional and behavioural interventions. Our admissions team is also able to use the engine to help with placement of new students throughout the year. Because we track languages spoken by students across the grade, our admissions team is also able to use the engine to help place new students starting in the middle of the year, seeking to place them with other students who speak their mother language.

These are just some of the positive outcomes that have arisen as a result of the design of this engine. For information on designing best-fit classes with the placement engine, feel free to contact Ben Hacking bhacking@vis.ac.at

Comments

Post a Comment

Popular posts from this blog

Using Social and Emotional Data to Promote a Positive School Culture

I came across this great article on Edutopia by Victoria Curry and Mike Setaro on how school leaders can combine traditional data with social and emotional data to get a full picture of the school experience of students and staff. It’s centered around Warm data that gives both dimension and measure to an individual’s and group’s social and emotional status. Opposite to Cool data points, that are a series of structural data sets such as enrollment, attendance, and academic proficiency that typically are the bedrock of school-based analytics. They talk about various examples of Warm data points among them on a matrix with different degrees of pleasantness and energy before engaging in learning (inspired by Mark Brackett’s work). These points can and should be captured, measured and visualized. The insights from this data should be of utmost importance for leaders to find strategies that capture and leverage information related to SEL and interpersonal skills. Harnessing this level of und...
Post a Comment
Read more

Atlas Protocol - Looking at data

The Atlas Protocol is one of the most commonly used structured dialogue formats among schools to facilitate a conversation about data for teachers and other members of the faculty. Learning from Data is a tool to guide groups of teachers discovering what students, educators, and the public understands and how they are thinking. The tool, developed by Eric Buchovecky, is based in part on the work of the Leadership for Urban Mathematics Project and of the Assessment Communities of Teachers Project. The tool also draws on the work of Steve Seidel and Evangeline Harris-Stefanakis of Project Zero at Harvard University. The protocol gives a detailed step by step guide on how to prepare and conduct a healthy, productive conversation about and with use of data. It starts with a selection of datasets that do not lead to a single conclusion and generally lead to rich conversations. From that point forward, the protocol describes 6 stages the group has to follow with the help of a facilitator in...
Post a Comment
Read more

What Data Can’t Do

I came across this fascinating article from the The New Yorker that really speaks volumes about how careful we have to be when it comes to using and visualizing data. Whenever you try to force the real world to do something that can be counted, unintended consequences abound. The COVID pandemic demonstrated just how vulnerable the world can be when you don’t have good statistics, and the US Presidential election filled our newspapers with polls and projections, all meant to slake our thirst for insight. The same applies to education where knowing what to measure, but also why you want to measure it, is the primary hurdle to tackle. We all have a tendency to naturally trust data as it aims to represent something we are observing. However there are times when simply even solid data is not enough for decision making. That’s why the context, the aim, and the balance between quantitative and qualitative data is so important. As the article states: “The great psychologist Daniel Kahneman, w...
Post a Comment
Read more